??? wrote: > CREATE TABLE t(x INTEGER PRIMARY KEY ASC, y, z); > > In order to testing the efficiency. After I create index on t(x). > And this give me a suprise, now I retrieve by x , I found that its is > faster than its before.
An index entry is smaller than a table row, so if you want to read only the values in the x column, searching in the index is likely to be faster then searching the rowid in the table, because less I/O needs to be done. This is called a covering index. Regards, Clemens